A method for completing coal wall point cloud in fully mechanized working face based on residual optimization
-
摘要: 煤矿综采工作面巷道的数字化三维重建过程中需要完整且密集的煤壁点云数据。受遮挡、视角限制等因素影响，采集的综采工作面煤壁点云数据往往不完整且稀疏，影响下游任务，需进行煤壁点云修复和补全。目前缺少针对井下点云补全任务的数据集和网络模型，现有模型用于煤壁点云补全时存在点云密度分布不均匀、点云特征信息丢失等情况。针对上述问题，设计了一种基于残差优化的煤壁点云补全网络模型，采用监督学习方式学习点云特征信息，通过最小化密度采样和残差网络迭代优化输出完整点云。采集煤矿井下真实综采工作面煤壁点云数据，预处理后筛选可用数据，通过模拟随机空洞制作煤壁点云缺失数据集，并用缺失数据集训练基于残差优化的煤壁点云补全网络模型。实验结果表明：与经典的FoldingNet，TopNet，AtlasNet，PCN，3D−Capsule点云补全网络模型相比，基于残差优化的煤壁点云补全网络模型针对构造的缺失煤壁点云和稀疏煤壁点云补全的倒角距离、地移距离及F1分数均能达到最优水平，整体补全效果最佳；针对实际缺失的煤壁点云，该模型能够实现有效补全。Abstract: The digital 3D reconstruction process of coal mine fully mechanized working face roadways requires complete and dense coal wall point cloud data. Due to factors such as occlusion and limited viewing angle, the collected coal wall point cloud data of the fully mechanized working face is often incomplete and sparse, which affects downstream tasks and requires coal wall point cloud repair and completion. At present, there is a lack of datasets and network models for underground point cloud completion tasks. Existing models used for coal wall point cloud completion suffer from uneven distribution of point cloud density and loss of point cloud feature information. In order to solve the above problems, a coal wall point cloud completion network model based on residual optimization is designed. Supervised learning is used to learn point cloud feature information, and the complete point cloud is output by minimizing density sampling and iteratively optimizing the residual network. The method collects real coal wall point cloud data of fully mechanized working face underground, preprocesses and screens available data. The method creates a coal wall point cloud missing dataset by simulating random cavities. The missing dataset is used to train the residual optimization-based coal wall point cloud complementary network model. The experimental results show that compared with the classic FoldingNet, TopNet, AtlasNet, PCN, and 3D-Capsule point cloud completion network models, the residual optimization-based coal wall point cloud completion network model achieves the optimal level of chamfer distance, ground shift distance, and F1 score for the constructed missing and sparse coal wall point clouds, with the best overall completion effect. It is able to achieve effective completion for the actual missing coal wall point clouds.
-
图 1 综采工作面煤壁点云获取过程
Figure 1. Acquisition process of coal wall point cloud in fully mechanized working face
图 2 煤壁点云预处理及缺失数据集构造过程
Figure 2. Preprocessing process of coal wall point cloud and creating process of missing dataset
图 4 基于残差优化的煤壁点云补全网络模型结构
Figure 4. Structure of coal wall point cloud completion network model based on residual optimization
图 7 不同网络模型对缺失煤壁点云补全效果的定量分析
Figure 7. Quantitative analysis of completion effect of different network models on missing coal wall point clouds
图 8 不同网络模型的煤壁点云补全可视化效果
Figure 8. Visualization effect of coal wall point cloud completion of different network models
图 9 不同网络模型对稀疏煤壁点云补全结果的定量分析
Figure 9. Quantitative analysis of completion effect of different network models on sparse coal wall point clouds
图 10 真实缺失煤壁点云补全定性分析
Figure 10. Qualitative analysis of coal wall point cloud completion in the case of true absence
表 1 激光雷达参数
Table 1. LiDAR parameters
参数 值 参数 值 角分辨率/（°） 0.25 输入电压/V 12/24 扫描频率/Hz 40 数据接口 EIP协议 功率/W 15 可输出点云格式 ply，pcd 
下载: 导出CSV
表 2 43101综采工作面点云坐标范围
Table 2. Point cloud coordinate range of 43101 fully mechanized working face
点云坐标 最小值/m 最大值/m X 6.982 7 348.288 8 Y −3.109 1 3.502 9 Z −0.005 0 3.102 4
下载: 导出CSV
表 3 实验环境硬件配置
Table 3. Hardware configuration of experimental environment
名称 配置 处理器 Intel（R） Xeon（R） CPUE5−2630 CPU主频 2.20 GHz 显卡 NVIDIA GeForce RTX 2080Ti 内存 256 GiB SSD 显存 16 GiB
下载: 导出CSV
表 4 实验环境软件配置
Table 4. Software configuration of experimental environment
名称 配置 操作系统 Ubuntu18.04 深度学习框架版本 Pytorch1.10.2 CUDA版本 10.0 CUDNN 版本 7.6.4 开发语言版本 Python3.8.11
下载: 导出CSV
-
[1] 王国法,刘峰,庞义辉,等. 煤矿智能化——煤炭工业高质量发展的核心技术支撑[J]. 煤炭学报,2019,44(2):349-357.WANG Guofa,LIU Feng,PANG Yihui,et al. Coal mine intellectualization:the core technology of high quality development[J]. Journal of China Coal Society,2019,44(2):349-357. [2] 郭金刚,李化敏,王祖洸,等. 综采工作面智能化开采路径及关键技术[J]. 煤炭科学技术,2021,49(1):128-138.GUO Jingang,LI Huamin,WANG Zuguang,et al. Path and key technologies of intelligent mining in fully-mechanized coal mining face[J]. Coal Science and Technology,2021,49(1):128-138. [3] 毛善君,鲁守明,李存禄,等. 基于精确大地坐标的煤矿透明化智能综采工作面自适应割煤关键技术研究及系统应用[J]. 煤炭学报,2022,47(1):515-526.MAO Shanjun,LU Shouming,LI Cunlu,et al. Key technologies and system of adaptive coal cutting in transparent intelligent fully mechanized coal mining face based on precisegeodetic coordinates[J]. Journal of China Coal Society,2022,47(1):515-526. [4] 李首滨,李森,张守祥,等. 综采工作面智能感知与智能控制关键技术与应用[J]. 煤炭科学技术,2021,49(4):28-39.LI Shoubin,LI Sen,ZHANG Shouxiang,et al. Key technology and application of intelligent perception and intelligent control in fully mechanized mining face[J]. Coal Science and Technology,2021,49(4):28-39. [5] XING Zhizhong,ZHAO Shuanfeng,GUO Wei,et al. Processing laser point cloud in fully mechanized mining face based on DGCNN[J]. ISPRS International Journal of Geo-Information,2021,10(7). DOI: 10.3390/IJGI10070482. [6] SI Lei,WANG Zhongbin,LIU Peng,et al. A novel coal-rock recognition method for coal mining working face based on laser point cloud data[J]. IEEE Transactions on Instrumentation and Measurement,2021,70:1-18. [7] 杨俊哲,姜龙飞,李梅,等. 基于激光点云的掘进工作面三维场景重建技术研究[J]. 煤炭科学技术,2021,49(增刊1):40-45.YANG Junzhe,JIANG Longfei,LI Mei,et al. Research on extraction technology of coal wall and roof boundary based on laser point cloud[J]. Coal Science and Technology,2021,49(S1):40-45. [8] XING Zhizhong,ZHAO Shuanfeng,GUO Wei,et al. Analyzing point cloud of coal mining process in much dust environment based on dynamic graph convolution neural network[J]. Environmental Science and Pollution Research,2023,30(2):4044-4061. doi: 10.1007/s11356-022-22490-2 [9] 王家臣. 我国综放开采40年及展望[J]. 煤炭学报,2023,48(1):83-99.WANG Jiachen. 40 years development and prospect of longwall top coal caving in China[J]. Journal of China Coal Society,2023,48(1):83-99. [10] XU Shaoyi,SHI Boxuan,WANG Chengtao,et al. Novel high-performance automatic removal method of interference points for point cloud data in coal mine roadway environment[J]. International Journal of Remote Sensing,2023,44(5):1433-1459. doi: 10.1080/01431161.2023.2184215 [11] 荣耀,曹琼,安晓宇,等. 综采工作面三维激光扫描建模关键技术研究[J]. 工矿自动化,2022,48(10):82-87.RONG Yao,CAO Qiong,AN Xiaoyu,et al. Research on key technologies of 3D laser scanning modeling in fully mechanized working face[J]. Journal of Mine Automation,2022,48(10):82-87. [12] 罗开乾,朱江平,张建伟. 三维点云补全方法的现状和发展趋势[J]. 信息记录材料,2020,21(5):179-180.LUO Kaiqian,ZHU Jiangping,ZHANG Jianwei. Current situation and development trend of 3D point cloud completion method[J]. Information Recording Materials,2020,21(5):179-180. [13] BERGER M,TAGLIASACCHI A,SEVERSKY L,et al. State of the art in surface reconstruction from point clouds[C]. 35th Annual Conference of the European Association for Computer Graphics,Strasbourg,2014:7-11. [14] KROEMER O,AMOR H B,EWERTON M,et al. Point cloud completion using extrusions[C]. 12th IEEE-RAS International Conference on Humanoid Robots,Humanoids,2012:680-685. [15] HANE C,SAVINOV N,POLLEFEYS M. Class specific 3D object shape priors using surface normals[C]. The IEEE Conference on Computer Vision and Pattern Recognition,Columbus,2014:652-659. [16] LI Yangyan,DAI A,GUIBAS L,et al. Database-assisted object retrieval for real-time 3D reconstruction[J]. Computer Graphics Forum,2015,34(2):435-446. doi: 10.1111/cgf.12573 [17] 刘彩霞,魏明强,郭延文. 基于深度学习的三维点云修复技术综述[J]. 计算机辅助设计与图形学学报,2021,33(12):1936-1952.LIU Caixia,WEI Mingqiang,GUO Yanwen. 3D point cloud restoration via deep learning:a comprehensive survey[J]. Journal of Computer-Aided Design & Computer Graphics,2021,33(12):1936-1952. [18] 刘心溥,马燕新,许可,等. 嵌入Transformer结构的多尺度点云补全[J]. 中国图象图形学报,2022,27(2):538-549.LIU Xinpu,MA Yanxin,XU Ke,et al. Multi-scale transformer based point cloud completion network[J]. Journal of Image and Graphics,2022,27(2):538-549. [19] 曾伟平,陈俊洪,ASIM M,等. 基于多阶段分形组合的点云补全算法[J]. 计算机与现代化,2023(12):24-29.ZENG Weiping,CHEN Junhong,ASIM M,et al. Point cloud completion algorithm based on multi-stage fractal combination[J]. Computer and Modernization,2023(12):24-29. [20] 陆春媚,杨志景. 多级精细化反卷积点云补全网络[J]. 计算机工程与应用,2023,59(17):242-249. doi: 10.3778/j.issn.1002-8331.2205-0500LU Chunmei,YANG Zhijing. Multistage refinement of deconvolution point cloud complementation network[J]. Computer Engineering and Applications,2023,59(17):242-249. doi: 10.3778/j.issn.1002-8331.2205-0500 [21] 王海军,刘再斌,雷晓荣,等. 煤矿巷道三维激光扫描关键技术及工程实践[J]. 煤田地质与勘探,2022,50(1):109-117. doi: 10.12363/issn.1001-1986.21.10.0589WANG Haijun,LIU Zaibin,LEI Xiaorong,et al. Key technologies and engineering practice of 3D laser scanning in coal mine roadways[J]. Coal Geology & Exploration,2022,50(1):109-117. doi: 10.12363/issn.1001-1986.21.10.0589 [22] 康传利,时满星,陈洋,等. 一种考虑多尺度噪声的平滑去噪方法[J]. 科学技术与工程,2018,18(11):110-116. doi: 10.3969/j.issn.1671-1815.2018.11.017KANG Chuanli,SHI Manxing,CHEN Yang,et al. A smoothing de-noising method considering multi-scale noise[J]. Science Technology and Engineering,2018,18(11):110-116. doi: 10.3969/j.issn.1671-1815.2018.11.017 [23] GROUEIX T,FISHER M,KIM V G,et al. A papier-mâché approach to learning 3D surface generation[C]. The IEEE Conference on Computer Vision and Pattern Recognition,Salt Lake City,2018:216-224. [24] YANG Yaoqing,FENG Chen,SHEN Yiru,et al. FoldingNet:point cloud auto-encoder via deep grid deformation[C]. The IEEE Conference on Computer Vision and Pattern Recognition,Salt Lake City,2018:206-215. [25] TCHAPMI L P,KOSARAJU V,REZATOFIGHI H,et al. TopNet:structural point cloud decoder[C]. The IEEE/CVF Conference on Computer Vision and Pattern Recognition,Seoul,2019:383-392. [26] YUAN Wentao,KHOT T,HELD D,et al. PCN:point completion network[C]. International Conference on 3D Vision,Verone,2018:728-737. [27] ZHAO Yongheng,BIRDAL T,DENG Haowen,et al. 3D point capsule networks[C]. The IEEE/CVF Conference on Computer Vision and Pattern Recognition,Seoul,2019:1009-1018. -
点击查看大图
计量
- 文章访问数: 82
- HTML全文浏览量: 34
- PDF下载量: 11
- 被引次数: 0
